Dynamic Walking Gait Designing for Biped Robot Based on Particle Swarm Optimization

Abstract

Versatility and energy cost of bipedal walking largely depends on reference gaits of robot. To obtain optimal reference gaits which are diversified in walking speed, an optimization method is proposed based on multi-object Particle Swarm Optimization. By rewriting the domination definition of multi-object optimization, the proposed method is able to get low energy consumption gaits of diverse walking speeds. Simulation results show that the designed optimization method is feasible and effective.